GEORGE SCHMIDT
The MIT Faculty has made this article openly available. Please share
how this access benefits you. Your story matters.
Citation
“People in Control [Interviews with George Schmidt and
Wolfgang Marquardt].” Control Systems Magazine, IEEE 29.5
(2009): 31-35. © 2009 Institute of Electrical and Electronics
Engineers.
As Published
http://dx.doi.org/10.1109/mcs.2009.933525
Publisher
Institute of Electrical and Electronics Engineers
Version
Final published version
Accessed
Wed May 25 18:19:23 EDT 2016
Citable Link
http://hdl.handle.net/1721.1/55976
Terms of Use
Article is made available in accordance with the publisher's policy
and may be subject to US copyright law. Please refer to the
publisher's site for terms of use.
Detailed Terms
PEOPLE IN CONTROL
I
n this issue of IEEE Control Systems Magazine (CSM) we speak with George
Schmidt, who is a lecturer in aeronautics and astronautics at MIT and an industry consultant in guidance, navigation, and control. George has worked on control
system design for missiles, aircraft, and manned spacecraft, Kalman filtering applications, and integration techniques for high-resolution synthetic aperture radars,
satellite navigation systems, and inertial sensors. He is an AIAA Fellow and an
IEEE Life Fellow, and he has been editor-in-chief of the AIAA Journal of Guidance,
Control, and Dynamics since 1996.
We also speak with Wolfgang Marquardt of the RWTH University Aachen, Germany. Wolfgang is an IFAC fellow, and he is editor-in-chief of the Journal of Process
Control. His expertise is in chemical process systems, process operations and control, model-based experimental analysis of chemical process systems, and numerical methods for dynamical simulation and optimization.
GEORGE SCHMIDT
Q.
Can you briefly describe your
background in aerospace engineering.
George: I grew up in New Jersey
near Newark and was always fascinated by the airplanes taking off and
landing at the airport. When I was a
sophomore in high school, Sputnik 1
was launched, and that inspired me
to become a member of the aerospace
profession. In 1960 I went to MIT to
study aeronautics and astronautics,
eventually receiving the S.B. and
S.M. degrees in 1965 and the Sc.D. in
instrumentation in 1971. But my professional career really started in 1961
when I started working at the Instrumentation Laboratory (the I-Lab),
part of the Department of Aeronautics and Astronautics. The I-Lab was
led by the head of the department,
Charles Stark Draper, who has been
called the father of inertial guidance.
The I-Lab seemed like a good place
to work part time as it was only a
block away from the MIT classrooms,
Digital Object Identifier 10.1109/MCS.2009.933525
and guidance and navigation based
on inertial navigation sounded like
magic. So I went to see Doc Draper
about getting a job. “Sure” he said.
“Go see Eleanor in Personnel.” So
began my career. And in that year,
1961, the I-Lab was awarded the first
NASA contract in the United States
to start work on the Apollo Program
to develop the guidance, navigation,
and control (GNC) system for the
spacecraft. I was very lucky to be
in the right place at the right time.
The I-Lab was divested by MIT in
George Schmidt (right) with his son Jonathan P. Schmidt and grandson Jonathan
G. Schmidt at Lake Winnepesaukee, New
Hampshire.
«
1973 and renamed the Charles Stark
Draper Laboratory. In 2007, I retired
after 46 years of continuous service.
Q. What kinds of projects were you
involved in at Draper Labs?
George: The 1960s at the I-Lab were
fantastic. In 1965, I completed my S.M.
thesis research on the application of
Kalman filtering for pre-launch alignment and calibration of the Apollo inertial measurement unit. I then became a
member of the technical staff and actually implemented my research into the
onboard Apollo guidance computer. In
1966 the algorithms began to be routinely used in the spacecraft. A 13-state
filter was employed operating in a
slow onboard computer. This was the
first application of Kalman filtering to
a production inertial system. Best of
all, I actually got to train many of the
Apollo astronauts when they visited
Cambridge, and I worked in the spacecraft control room at Cape Kennedy
for every flight through Apollo 13.
When I finished my Sc.D. thesis in
1971, I decided I needed to learn more
about frequency-domain problems so
I transferred from NASA programs
to work on Air Force programs. I was
asked to work on the motion compensation needed for implementation
of synthetic aperture radars (SAR) in
high-speed platforms. Well, it turned
out to be another time-domain problem with filtering involved, so I hadn’t
escaped at all! But it was tremendously
interesting and eventually involved
B-52 flight testing. Nowadays, SAR is
commonly implemented in strategic
bombers and tactical aircraft.
Late in the 1970s, I got a chance
to work on defining the avionics
OCTOBER 2009
« IEEE CONTROL SYSTEMS MAGAZINE
Authorized licensed use limited to: MIT Libraries. Downloaded on November 16, 2009 at 15:13 from IEEE Xplore. Restrictions apply.
31
George Schmidt with his granddaughters
Kathryn (left) and Alyssa (right).
requirements to launch TERCOMequipped cruise missiles from widebody aircraft. The program never progressed, but I learned about strategic
aircraft and strategic cruise missiles. I
proposed a study to the Air Force that
would define the avionics requirements for future strategic aircraft and
strategic cruise missiles that would
be significantly less dependent on
TERCOM. I am pleased to say that
today many of the key study recommendations were implemented in the
unique B-2 navigation system and
also in the stealthy Advanced Cruise
Missile (ACM). Four hundred sixty
one ACMs were produced and are
now in the process of being retired
from service. Fortunately, these
nuclear-equipped, stealthy missiles
were never used, but they were part
of the strong deterrent we presented
to the Soviet Union.
Throughout the 1970s and into
the 1990s, global positioning satellite
(GPS) development and applications
were proceeding at a furious pace. The
Defense Department was beginning to
understand that the GPS signal was
fragile in the sense that a determined
opponent could jam the received signal. I was fortunate to serve on several
panels, including the Defense Science
Board Task Force on GPS, that looked
at ways to preserve our ability to use
GPS in a hostile environment while
denying an enemy’s ability to use
GPS. As a result of that study, a large
national program, NAVWAR, was
started that developed many improvements to address future military
needs. One of those improvements
32 IEEE CONTROL SYSTEMS MAGAZINE
»
was to develop advanced inertial system/GPS integration techniques to
increase the antijam capability of the
system. Since inertial systems cannot
be jammed, they offer the ability to
provide accurate navigation throughout a period of GPS jamming.
In 1998 I became the education
director of Draper Laboratory. As
part of the 1973 MIT divestiture of the
I-Lab, Draper Laboratory continued to
have MIT graduate students do their
thesis research at the laboratory. From
1998 through 2007 when I retired,
each year we had about 60 graduate
students from MIT doing research. In
addition, I was directing a US$2 million per year university research program with typically 18 projects a year
at MIT and universities throughout
the country.
When I retired from Draper Laboratory, I felt I accomplished my goal
to be a successful aerospace engineer,
that I contributed to nationally important projects, and that I also contributed to developing the next generation
of technical leaders of our nation. Now
I am concentrating on teaching, consulting, and my AIAA and IEEE professional activities.
Q.
In 2005 you were awarded the
NATO Research and Technology Organization’s highest technical award, the
von Kármán Medal. What contributions led to that specific recognition?
George: Beginning in 1968, I started
to contribute to the NATO research
organization technical programs as
a paper presenter at conferences, as
a research study group member, as a
conference organizer, and as a lecture
series director. I believe I hold the
record for organizing the most NATO
lecture series focused on specific topics of NATO interest. After the fall
of the Iron Curtain, I organized the
first lecture series in Russia. I also
used my NATO connections to find
international advisors for the Journal
of Guidance, Control, and Dynamics.
At the time of the award in 2005, the
contributions to NATO international
programs had spanned 37 years, and
that body of work was the basis of
the medal award, which states “He
has been a catalyst for international
cooperation for the common good
of all mankind.” I am still involved
with the NATO Research and Technology Organization (RTO) and will
be directing a lecture series, “LowCost Navigation Sensors and Integration Technology,” in Kiev, Prague,
Toulouse, and Lisbon in March 2010.
This same lecture series was presented
in Madrid, the Delft, and Farnborough in 2008 and in Rome, Munich,
and Warsaw in 2009. Under the IEEE
Aerospace and Electronics Systems
Society Distinguished Lecturers
Program, I will be giving lectures in
Melbourne, Adelaide, and Brisbane
in October 2009. I enjoy international
travel and meeting engineers with
different backgrounds.
Q.
You’re active in the NATO Research and Technology Organization,
which used to be called AGARD.
Can you describe the kinds of
control-related technology that this
group is involved in.
George: The NATO Advisory Group
for Aerospace Research and Development (AGARD) was formed by von
Kármán in 1952. The NATO Defense
Research Group (DRG) was formed
later, and about ten years ago they
were combined as the NATO RTO.
RTO promotes and conducts cooperative scientific research and exchange
of technical information among 28
NATO nations and 38 NATO partners.
The effort is managed by an executive agency, RTA, which facilitates
collaboration by organizing a wide
range of studies, workshops, symposia,
lecture series, and other forums in which
researchers can meet and exchange
knowledge. Actual research activities
are supported by the member nations
themselves, not by RTO, so research
contracts are not issued by RTO. Six
RTO panels are actively involved in
conducting these cooperative activities. A complete list of activities and
publications is given at www.rto.nato.
int. I believe control is an enabling
OCTOBER 2009
Authorized licensed use limited to: MIT Libraries. Downloaded on November 16, 2009 at 15:13 from IEEE Xplore. Restrictions apply.
technology that can be found in virtually all of the RTO activities listed on
the Web site, which includes the complete scope of defense technologies.
Q. What do you view as some of the
key current challenges in aerospace
engineering related to guidance and
control?
George: I believe we are on the path
to achieving worldwide 1-m combined
INS/GPS navigation accuracy at low
cost with robustness to interference.
INS systems will be based on MEMS
technology, so they will be small, lightweight, and use little power. Improvements in the accuracy of GPS are
already nearing that accuracy goal,
and so is the INS/GPS antijam capability. However, the greatest challenge
is improving MEMS gyros and accelerometers by two orders of magnitude in
performance. If that can be achieved,
we will have small size, low cost,
and high performance. The uses of
INS/GPS will proliferate. Individuals,
including civilians and soldiers, will
have personal navigators, guided artillery shells will have superb accuracy,
and robots will be able to accomplish
un-thought-of missions. Other applications will surely include spacecraft,
aircraft, missiles, commercial vehicles,
and consumer items. The second
challenge will be in preparing maps
that have better than 1-m accuracy in
which the INS/GPS will navigate.
There is a third challenge, which is
to harness the increasing power of
onboard computers. Implementing
algorithms such as nonlinear filters is
becoming possible now even though
they were dismissed as impractical a
few decades ago.
Q. The AIAA Journal of Guidance,
Control, and Dynamics (JGCD) is
one of the premiere control-related
journals in the aerospace field. Congratulations on your “guidance” of
that publication since 1996. Modestly speaking, of course, how have
you managed to bring the publica-
tion to its current state of distinction? Also, what advice do you have
for authors who wish to submit articles for publication?
George: Thank you for the kind
words. The JGCD was started in 1978
by Donald Fraser, and under his leadership for 14 years it rose to a high status.
Terry Alfriend continued that stewardship until 1996. When I started, the
JGCD was in excellent shape and well
beyond any startup transient effect.
More than any individual, the 23
associate editors (AEs) control the
quality of the journal, and that is
what I try to stress when I recruit
them. Selecting the AEs is my most
important job. I regularly update
my list of outstanding reviewers for
potential AEs. The JGCD has been
lucky to have three AEs serve at least
as long as I have served. The retirement rate of AEs, who serve threeyear terms, is very low, typically
three per year. That allows a very
real continuity in review quality in
our technical areas. I try to balance
the number of AEs equally among
academia, industry, and government
laboratories. The JGCD is supposed
to be an applications-oriented journal so it is important to balance the
different AE backgrounds.
As for author advice:
1) Read all the way through the submission requirements on the Web
and then follow them exactly.
2) Clearly identify your contribution and its relationship to prior
work as documented by references preferably chosen from
peer-reviewed publications.
3) Make sure all sentences are
grammatically correct and make
sense technically! Non-English
speaking authors should note
that they only get three chances
to submit in grammatically correct English. Reviewers will not
review papers when they have to
correct the grammar.
4) If your paper is a bunch of equations followed by some trivial,
simulated aerospace example,
don’t submit. Revise it to show
you can solve some practical aerospace problem.
Q. You were education director at
Draper Labs. What trends do you see
in control-related education?
George: I still teach a graduate
course at MIT, and one positive trend
that is obvious to me is that there is
now greater collaboration between
faculty, even in different departments.
For example, control, communications, software, and other faculty are
conducting interesting systems-level
research involving graduate students
in transportation systems, satellite
constellation design, as well as in
bio-inspired navigation. The theses
produced by these students are truly
amazing. My only complaint is that it
seems as if mathematics is a foreign
language to some students. I don’t
know what the cause is, but I am convinced that many students cannot do
the rigorous analytical work required
in graduate school. Can it be the
over-dependence on computers and
canned analysis programs? But my
major concern is the decreasing financial support by the government and
industry that is leading to a future
disaster in the number and quality of
engineers we graduate.
Q. What are your interests outside
of professional activities?
George: I have two married sons
and three grandchildren under two
years old. Fortunately, I see and provide day care to two of them often. I
have a house in New Hampshire on a
lake and enjoy boating in the summer
and skiing in the winter. I also like
to travel, preferably internationally.
And I have the best long-term companion, Judy, whom I could imagine
or dream about.
Q.
Thank you for speaking with
CSM!
George: You’re most welcome.
OCTOBER 2009
«
IEEE CONTROL SYSTEMS MAGAZINE 33
Authorized licensed use limited to: MIT Libraries. Downloaded on November 16, 2009 at 15:13 from IEEE Xplore. Restrictions apply.
WOLFGANG MARQUARDT
operational strategy or the startup or
shutdown of a unit in a plant.
In recent years, we have developed
powerful optimal control algorithms
that aim at a fully adaptive solution
strategy of the dynamic optimization
problems to achieve robustness as well
as computational efficiency in a fully
nonlinear setting. Furthermore, we
always emphasize a late discretization
approach to fully exploit the power
of numerical algorithms rather than
looking at the discrete-time problem.
More recently, we have worked on
neighboring extremal control to
reduce the computational effort and
to develop consistent decomposition
strategies in a hierarchical approach.
Most recently, we’ve been interested
in decentralization aspects to deal
with functional decomposition of large
plants either to reduce complexity of
the optimization scheme or to deal
with different time-scale processes.
Modeling and model tuning including
integrated parameter and state estimation to fight the inherent process and
model uncertainty becomes the true
challenge if these control systems are
applied to a real industrial process.
to effectively cope with the network
character of such systems. Only partial coordination of the subsystems
of different stakeholders is possible.
The models are strongly nonlinear,
large scale, and typically characterized by widely varying time scales.
The theoretical foundations and the
practical issues of engineering and
implementing such highly structured
optimization-based systems in an
industrial environment are largely
open. The investigation of these kinds
of problems, however, requires us to
somehow abandon the traditional
view of control systems engineering where the control objective is assumed to be given as a setpoint and
some desired control performance.
Rather, control has to take a broader
view. Its target should be the economic performance of a technical system
that is implemented by the plant, the
monitoring and automation system,
and the human operators and decision makers in face of process and
model uncertainty. The integration of
various design tasks, such as those related to the process or energy system
itself, the operational support system
comprising an automatic control and
optimization system, as well as the
human decision makers, is an obvious
challenge to tackle in such a setting.
Q. What are some of your research
projects in process control?
Wolfgang: The focus of our re search in process control is on the
development of methodologies and
algorithms for dynamic real-time
optimization (D-RTO) of process and
energy systems, such as a chemical
plant or a power plant, which are typically modeled by a large number of
nonlinear differential-algebraic equations. D-RTO is a variant of nonlinear
model-predictive control with an economical objective. Rather than tracking
a constant or time-varying setpoint
and attenuating fast disturbances in
the traditional setting in systems and
control, our goal is to operate the plant
at all times in an economically optimal
manner. In particular, the “control”
objective is the maximization of profit
during a finite-time horizon campaign
despite the time-varying nature of the
environment the plant is operating in.
The disturbances acting on the plant
include fast, typically random disturbances with zero mean as well as
slow disturbances with nonvanishing
trends on different time scales. These
can either be endogenous and a consequence of operator interaction, a grad- Q. What do you view as the most
ual deterioration of the performance challenging problems currently facof the equipment, exogenous resulting ing the field?
from ambient pressure and temperaWolfgang: Integrated optimizationture fluctuations, changes in the pro- based control systems will be imple- Q. How did you become interested
cessed raw materials, or in the desired mented in a complicated multiloop, in process control?
product qualities and quantities. Con- hierarchical, and decentralized archiWolfgang: I remember very well
sequently, setpoint generation as well tecture in process and energy systems a visit in the control laboratory of E.D.
as setpoint tracking are interGilles at Stuttgart University durtwined problems that are solved
ing my undergraduate studies in
either by an integrated approach
the first week of the mandatory
as in nonlinear model-predictive
introductory control course that
control or by a hierarchical dechemical engineering students
composition approach accounttake. He showed me a lab-scale
ing for the different time scales
chemical reactor operating in a
of real-time (reactive) producself-sustained oscillatory mode if
tion planning and scheduling,
no controller was active. He then
economical optimization of typi“tamed” the nonlinear dynamics
cally time-varying setpoints, and
of the reactor by means of a simof disturbance rejection and staple feedback controller. This simbilization. The decision variables
ple laboratory experiment and the
are not only the control variable
clear and enthusiastic explanasetpoint trajectories but are also Wolfgang Marquardt with his wife Barbara and their tion of the mechanisms underlyinteger variables encoding some daughter Vera visiting a castle in Paris.
ing both the open-loop nonlinear
34 IEEE CONTROL SYSTEMS MAGAZINE
»
OCTOBER 2009
Authorized licensed use limited to: MIT Libraries. Downloaded on November 16, 2009 at 15:13 from IEEE Xplore. Restrictions apply.
oscillations as well as feedback stabilization left a deep impression and led me
to major in systems and control.
Q. You are a recipient of the 2001
Leibniz-Award of the German Science
Foundation. Which of your contributions was honored by that award?
Wolfgang: Every year about ten
scientists across all the scientific disciplines are honored by the Leibniz
award of the German Science Foundation. The award was honoring me for
the development and application of
systems engineering and computer science methods to the modeling, design,
control, and operation of chemical
engineering processes.
Q. Please talk a little about the Journal of Process Control (IPC). How
would you describe the role of this
journal in the process control field?
Wolfgang: The objective of the JPC
is to publish papers on the theory and
applications in systems and control
that are of particular interest to the
energy and chemical process industries. Application papers have to go
beyond case studies and need to demonstrate novel solution concepts and
strategies for a broad class of problems.
Industrial implementations of frontier
control technologies are also of interest. Theoretical papers constitute an
important part of the portfolio of JPC.
They have to relate to applications,
however. Consequently, such contributions not only have to be original and
correct but have to address a problem
that addresses a challenging problem
in applications. JPC has become the
dissemination platform for the latest
research results in process control and
related fields. The portfolio of papers
published vividly demonstrates the
high quality of research and activity of
the process control community.
Q. Do you see any trends in undergraduate and graduate education
relating to process control, especially
at RWTH University Aachen?
Wolfgang: The undergraduate education of chemical and mechanical
The caberet group VT5, from left: Thomas Melin, Jochen Büchs, Michael Modigell, Wolfgang Marquardt, Andreas Pfennig, and Mirka Mörl (photo courtesy of Hendrik Brixius).
engineers in control should in the first
place emphasize modeling and simulation in the context of dynamical systems theory without introducing the
concept of feedback in the very beginning. This way, the students can relate
the abstract setting of systems and
control to the engineering fundamentals in mechanics and thermodynamics. The state space is a natural setting
directly resulting from their modeling
experience in the mechanics and thermodynamics courses. Then, feedback
and control system design can gradually be introduced, ideally integrating
time-domain and frequency-domain
methods. Systems and control has to
be understood and taught as an engineering science rather than a topic in
applied mathematics without losing
the sound theoretical basis.
Unfortunately, in many universities—including ours—the students
largely consider the control course a
hurdle to be overcome on their way to
the degree rather than a fascinating and
relevant topic for practicing engineers.
It is this mindset that any undergraduate education in control has to change
to get a higher proficiency of practicing
engineers in systems and control engineering. I consider this more important
than optimizing the curriculum for the
systems and control experts. For these
students, there is not just a single road
to heaven. We have much more interesting and relevant material to teach
than the students can digest. A selection with a focus on the fundamen-
tal concepts and relevant application
classes is required. This selection can
and always should reflect the special
research expertise of the instructor—
following Humboldt’s principle of the
unity of education and research.
Q. What courses do you especially
like to teach?
Wolfgang: I like most to teach a
systems-theory-centered introduction
to modeling and simulation for undergraduate students, advanced modeling
of chemical and energy process systems,
and applied numerical optimization.
Q. What are some of your hobbies
and interests outside of professional
activities?
Wolfgang: I like to listen to all
kinds of music. But I am also practicing music. For some years now,
four of my chemical engineering colleagues and I have formed a cast (VT5)
practicing and performing a cappella
cabaret. With our songs and presentation we make fun of ourselves, our
profession, the university, and the
political and societal system. We have
had a number of performances at
conference dinners and in the university to entertain colleagues and—in
particular—students. It’s been a lot of
fun and so different from what all of
us are doing professionally.
Q. Thank you for speaking with CSM!
Wolfgang: You’re most welcome.
OCTOBER 2009
«
IEEE CONTROL SYSTEMS MAGAZINE 35
Authorized licensed use limited to: MIT Libraries. Downloaded on November 16, 2009 at 15:13 from IEEE Xplore. Restrictions apply.